Forum: gecon

Policy Rules in gEcon [ Reply ]
By: Jesse Grabowski on 2021-08-13 13:13

Hello,

I am running into problems solving even small-scale DSGE models when fiscal policy rules are introduced. I have tried several functional forms, for example those suggested by Costa (2016).

What did work was following a form similar to the Taylor rule presented in Kilma et al. (2015), where a pair of calibrated parameters like calibr_pi and calibr_pi_obj are used to enforce a steady state relationship. This method of enforcing a steady-state value is not very intuitive to me, but regardless, similar policy functions have produced solutions in a medium-sized model (~100 equations). They were highly dependent on the initial values I passed to the solver, and broke immediately when perturbed, i.e. during calibration.

(As an aside, it would be very nice if more options, preferably all options, inside the bayesian_estimation function were exposed to the user, for example the tol term in function lyapunov, and init_draws in the metropolis-hastings sampler)

I suppose my question is, what advice do you have for writing effective policy rules in gEcon? In Dynare, it is possible to give specific steady state values, but this is not possible in gEcon (adding, for instance, pi[ss] = 1 causes it to bark at you for having too many equations). It seems one must rely on calibrated parameter tricks, but these are fragile and non-intuitive. Am I missing something very obvious?

Thank you,

Jesse